Universities must own their own AI – or risk irrelevance
Universities must own their own AI – or risk irrelevance
For years higher education has been told that artificial intelligence poses an existential threat. At first the predictions were easy to dismiss as hype. Universities have weathered technological revolutions before the rise of the internet, MOOCs, and now the proliferation of online course platforms. But AI is different. It is not just a tool for learning; it is a tool for doing. If students can summon personalised tutors, receive instant feedback, and generate publishable essays from their phones, what becomes of the traditional academic model? Add to this the erosion of the graduate premium as automation steadily eats into entry-level roles, and the threat is no longer hypothetical. It is immediate and real.
Already, graduate outcomes are faltering. In the UK, the Institute for Fiscal Studies has reported that the lifetime earnings advantage of a degree is narrowing, particularly for men outside elite institutions. Across the OECD, the proportion of graduates securing professional employment within six months of graduation is slipping. Employers increasingly automate routine entry-level jobs once filled by new graduates, ranging from paralegal work to financial analysis. These were precisely the stepping-stone roles that underpinned the value of a university education. As the graduate premium falls, so too does public confidence in higher education as a secure investment.
At the same time, the financial model of universities is under acute strain. In Britain, a decade-long freeze in domestic tuition fees has collided with double-digit inflation and a decline in international student numbers. The Office for Students estimates that a domestic student tuition fee of £9,250 is now worth closer to £6,000 in real terms. The result is widespread financial instability: nearly half of UK universities are forecast to be in the red within two years, particularly with the numbers of international students declining and with a growing number of UK universities dependent on commercial loans just to meet payroll.
In Australia, the Group of Eight universities have warned that government proposals for a new levy could strip hundreds of millions from research budgets. In both countries, international students have been the lifeline. Yet even this market is wobbling under visa restrictions, political hostility, and intensifying global competition. Against this backdrop, the arrival of AI appears to some as the final straw. If universities cannot guarantee strong graduate outcomes, cannot maintain financial stability, and cannot compete with cheaper, technology-enabled alternatives, why should they survive at all?
The temptation is to resist. Policymakers call for regulation, academic senates debate banning ChatGPT in assessments, and some even propose a pause on AI development. But this misunderstands the moment. AI is a genie that cannot be put back in the bottle. It is already embedded across finance, healthcare, logistics and government. Banks use AI to detect fraud in real time; hospitals use it to triage patients and manage capacity; governments deploy it to forecast tax revenues and benefits payments. No serious actor in any sector is waiting for permission to proceed.
Universities have no immunity. They can either adapt, or watch as others, private education providers, technology firms themselves, and new hybrid platforms define the future of education without them. The real question is not whether AI will be used in higher education, but rather, who will own it.
This is where universities must stop being passive consumers and start being proactive owners. Imagine a university with its own AI, built and trained on its own data, accountable to its own leadership and values. This would not be a futuristic gimmick but now an operational necessity. An autonomous university "AI Agent" could integrate staff and student data, estates management, financial systems, faculty and departmental budgets, research outputs, and even departmental reporting into a single, responsive system.
When I raised the idea with a vice-chancellor with a background in IT and systems, his response was cautious: “Great idea, but I doubt it will fly. Too many players involved. Who would own it?” In other words, a familiar litany of reasons to preserve the higher education status quo. But the status quo is unsustainable. Without change, universities risk being outflanked by nimbler competitors who understand that in a data-driven world, control of information is control of destiny.
Consider reporting. Today universities spend vast sums compiling data for regulators, funding bodies, and governments. Administrators wrestle with incompatible legacy systems — student records in one database, estates costs in another, staff payroll in a third. The duplication and inefficiency are substantial. According to Universities UK (2023), the regulatory and compliance burden in the higher education sector requires, on average, 17.6 full-time equivalent staff per institution, with total costs across the sector running into hundreds of millions of pounds annually.
Now imagine an AI system owned by the university itself. A regulator requests information on graduate employment by discipline: the AI generates it in seconds, without human teams cobbling together spreadsheets. A government agency asks for energy use and sustainability data: the AI provides a real-time dashboard. Boards of governors want clarity on research output or faculty workload: again, the AI delivers. Any individual within the institution could query the system, ending silo thinking and creating transparency across the board. The benefits are not just efficiency and cost-saving. They are transparency and trust. Universities would no longer appear defensive or opaque, but open, accountable, and proactive.
Universities pride themselves on autonomy. Yet ironically, the current trajectory risks giving it away. By outsourcing student systems, research platforms, and even assessment tools to external vendors, universities are steadily ceding control over their most valuable asset: their data. As Rick Morton recently reported in The Saturday Paper, the consultancy Nous Group has “slowly taken over the university sector, filling VCs’ offices with ex-staff and buying ‘incredibly sensitive’ university data that is sold back to them for benchmarking.” Once housed in a consulting or technology company’s servers, that data is no longer truly the university’s to direct. Dependence on external platforms may feel convenient today, but in the long run it erodes institutional independence and costs eyewatering sums to outsource.
Contrast this with healthcare, where many national health systems have developed their own AI for patient records and resource management precisely to retain sovereignty over sensitive information. Or finance, where banks maintain proprietary fraud-detection algorithms because outsourcing them would compromise security and competitiveness. Governments, too, are investing heavily in sovereign AI infrastructure to avoid dependence on big tech. Universities should ask themselves why they would treat their own intellectual and operational capital with less care than banks treat money or hospitals treat patients. Given the health and financial information regulated closely and held on AI, there is no reason university data cannot be housed on internal systems owned by universities themselves.
The message is clear. AI does not have to spell the end of the university. But if the sector continues to sit on the sidelines, outsourcing intelligence while defending inefficient legacy systems, then the dire predictions of decline may come true. In a world of shrinking graduate premiums, falling student numbers, and deepening financial deficits, doing nothing is no longer an option.
By building and owning their own AI, universities could reduce administrative burdens, increase transparency, strengthen their finances, and most importantly, preserve their autonomy. They would no longer be passive victims of technological disruption but active shapers of their own destiny. The genie is out of the bottle. The only question is whether universities will grasp it or whether they will look back a decade from now, halved in number, wondering why they hesitated when the future was already immediately within their reach.
